Download as pdf or txt
Download as pdf or txt
You are on page 1of 13

Skin Cancer Detector

Ahmed Nazmy, Ali Samy, Kerolos Samy, Mahmoud Said,

Abdalla Ahmed, Osama Khalil

idea

The main idea of this project is Automated Skin Cancer Detector.


Our system should provide a quick result with high accuracy. We
aim to detect several types of skin cancers.
Overview​ :

Skin cancers are ​cancers​ that arise from the ​skin​. They are due to the
development of abnormal ​cells​ that have the ability to invade or ​spread​ to
other parts of the body. There are three main types of skin cancers:
basal-cell skin cancer​ (BCC), ​squamous-cell skin cancer​ (SCC) and
melanoma​. The first two, along with a number of less common skin
cancers, are known as nonmelanoma skin cancer (NMSC). Basal-cell
cancer grows slowly and can damage the tissue around it but is unlikely to
spread to distant areas or result in death. It often appears as a painless
raised area of skin, that may be shiny with ​small blood vessels running over
it​ or may present as a raised area with an ​ulcer​. Squamous-cell skin cancer
is more likely to spread. It usually presents as a hard lump with a scaly top
but may also form an ulcer. Melanomas are the most aggressive. Signs
include a ​mole​ that has changed in size, shape, color, has irregular edges,
has more than one color, is itchy or bleeds.

Skin cancer is the most common form of cancer, globally accounting for at
least 40% of cases.The most common type is nonmelanoma skin cancer,
which occurs in at least 2-3 million people per year.
Of nonmelanoma skin cancers, about 80% are basal-cell cancers and 20%
squamous-cell skin cancers. Basal-cell and squamous-cell skin cancers
rarely result in death.

Globally in 2012 melanoma occurred in 232,000 people, and resulted in


55,000 deaths.
Problem Definition :

Skin cancer is ​the world’s most common cancer that strikes one in five people by
age 70. The good news is that 99 percent of all cases are curable if they are
diagnosed and treated early enough.

Early detection gives you the power to detect cancer early when it’s easiest to cure,
before it can become dangerous, disfiguring or deadly.

Skin cancer is the cancer you can see. Unlike cancers that develop inside the body,
skin cancers form on the outside and are usually visible.

Basal Cell Carcinomas​ are slow growing tumors that usually appear as smooth,
pink bumps with a pearly sheen that later crust, ulcerate, and bleed. They
commonly occur on the sun exposed parts of the body, especially the head, neck
and hands.

Squamous Cell Carcinomas​ are typically raised, scaly, pink, wart like growths,
which ulcerate as they enlarge. These are also common on the sun-exposed areas.
If not treated, this cancer can become a large mass and can spread to other parts of
the body.

The first sign of a ​Malignant Melanoma​ may be the development of a multicolored


mole with jagged uneven borders or any other change in size, shape or color.
Unlike the other skin cancers, these can occur on any part of the body. If not
treated at an early stage, these fast growing tumors can quickly spread to other
parts of the body.
Project idea :

Any device or service that can accurately give the probability of malignancy by
analyzing a simple photograph of the tumor would be very helpful for both primary
care doctors and their patients. In this context, the development of artificial
intelligence (AI) that can classify skin tumor images within seconds, at a skill level
similar to trained dermatologists, is an ideal solution for this problem.

Computer vision can play an important role in medical image diagnosis and it has
been proved by many existing systems.

The main goal of this project is to reduce skin cancer-related deaths by developing
a mobile application that can be used by any user and everywhere for early
detection of skin cancer, this application will reduce the cost of diagnosing and
treatment for patients and save lives.
Related work :

Several diagnostic systems for cancer detection have been proposed. Some systems
try to imitate the performance of dermatologists by detecting and extracting several
dermoscopic features. These features can then be used to score a lesion in a similar
way to the one adopted by dermatologists.

In [1], general clinical principles of early melanoma detection are reviewed,


providing the clinician with an up-to-date understanding of management strategies
for their patients with numerous or atypical nevi. Many researchers have been
working on the image processing and computer vision techniques for skin cancer
detection. The most probably features to perform skin lesion segmentation used in
various papers are shape, color, texture, and luminance.

Three methods of segmentation have been discussed by [2]. The methods are:
Otsu’s method, gradient vector flow (GVF), and color based using K-mean
clustering. Feature extraction is based on the so-called ABCD-rule of
dermatoscopy. While [3], a watershed segmentation is the proposed scheme used
for image segmentation, border detection and decision related with structural
nature of lesion.
In [4] Based on a qualitative assessment of asymmetry (of boundary, color, and
mass distribution), size functions (SFs) and support vector machine (SVM) are
used to implement a new automatic classifier of melanocytic lesions .

In [5] they applied a Bag-of-features approach to malignant melanoma detection


based on epiluminescence microscopy imaging (low-power microscopy (×50–100),
commonly a television microscope applied to a glass slide covering mineral oil on
the surface of a skin lesion, to determine malignancy in pigmented lesions). Each
skin lesion is represented by a histogram of code words or clusters identified from
a training data set. Classification results are achieved based on the implementation
of naive Bayes and support vector machine classifiers.

In [6] The feature extraction method is 2D wavelet transform, 2-D wavelet packet
is used and the enhanced image in gray scaled as an input. Bio wavelets at two
steps of decomposition are used. At each step of decomposition, the wavelet of
primary image is divided into an approximate and three detailed images which
show the basic information and vertical, horizontal and diagonal details,
respectively. The Features extracted using the wavelet transform are: Mean,
Standard deviation, Mean Absolute Deviation, L1 Norm, L2 Norm. Along with
these features, two additional features are taken – Skewness and Kurtosis.
Skewness is a measure of asymmetry and Kurtosis is a measure of whether the data
are peaked or flat relative to a normal distribution. .
These features are given as the input to the artificial neural network classifier
In [7] ​ ​they attached the Dermlite® DL1 dermatoscope to the iPhone. A new
method called elliptical symmetry was proposed for quantifying asymmetry.
Gaussian smoothing and lacunarity analysis to measure border irregularity were
proposed. In Gaussian smoothing, the contour was smoothed and compared with
the perimeter of the original lesion. The lacunarity was used to analysis the borders
of the image. Finally, the extracted features were fed to input layer of the
multi-stage neural network classifier.

In [8] Research recommended to use Using ANN to human ex- perts and
dermatologist who specializes in diagnosing and treatment of skin and related
diseases and who operate in areas where there are no specialist (dermatologist) can
also rely on the system for assistance.
The system achieved a high level of success using the artificial neural network
technique, with 90 percent success rate. This infers that ANN technique is an
efficient method for implementing diagnostic problems.
In [9] proposed a new fully automated enhanced deep CNN model for skin lesion
border identification and lesion recognition. The proposed heterogeneous
framework includes three major pipeline procedures including augmentation and
contrast enhancement, color pixels based on CNN training and segmentation, and
CNN features fusion and selection. The best subset of selected features is classified
through NN. A comparison result is conducted for each dataset using several
classification methods. The segmentation results are evaluated on PH2 and ISIC
2017 datasets and attained an average accuracy of 95.41% and 94.78%,
respectively. The proposed recognition process is evaluated on ISIB 2016, 17, and
PH2 datasets and attained the best accuracy of 95.1%, 94.8%, and 98.4%,
respectively. The proposed system accuracy is better as compared to the existing
state-of-the-art segmentation and recognition techniques.

In [10] The study on different techniques for skin cancer detection shows that
various techniques give fairly accurate results but they still left much to be desired.
The SVM model that followed show a substantial improvement in the accuracy of
skin cancer detection. By analysing the empirical results, we can see that
unsupervised learning algorithms like k-mean clustering and certain neural
networks like backpropogation neural networks have certain disadvantages, SVMs
are better at detection with a high degree of accuracy. A result having close to
perfect accuracy is yet to be seen due to factors like insufficient database and
failure of the proposed algorithms to achieve the ideal results. The study shows the
use of SVMs and CNNs will be able to provide promising results if the results can
be supplemented with additional intelligence
In [11] The methodology in this paper revolves around two aspects they are Digital
Image Processing and Convolutional Neural Network.

The cancer detection procedure consists of the following steps:


Image Acquisition - Conversion to HSV - Reduction of color channels -
Image Segmentation - Morphological Operations - Region of Interest.

The images obtained after the image processing procedure are fed to our CNN
model.

The proposed approach was tested on International Skin Imaging Collaboration


(ISIC) dataset, The image quality is some of the images in the dataset is not quite
good which naturally affects the accuracy of the classifier. The model achieved
accuracy of 74.76% and validation loss of 57.56%. As per the ISIC dermatologists
tend to get accuracy of 60% in identifying melanoma in a skin lesion without the
help of any dermoscopic imaging. Our model can help dermatologists in the early
detection of melanoma skin cancer up to a great extent.
References :

[1] Goodson AG, Grossman D. Strategies for early melanoma detection:


Approaches to the patient with nevi. Journal of the American Academy of
Dermatology. 2009;60(5):719-735. DOI: 10.1016/j.jaad.2008.10.065

[2] Bhuiyan MA, Azad I, Kal-Uddin M. Image processing for skin cancer features
extraction. International Journal of Scientific and Engineering Research.
2013;4(2):1-6

[3] Jain JW, Ramteke NS. ABCD rule based automatic computer-aided skin cancer
detection using MATLAB. International Journal on Computer Technology and
Applications. 2013;4(4):691-697

[4] D'Amico M, Stanganelli I. Qualitative asymmetry measure for melanoma


detection. In: IEEE International Symposium on Biomedical Imaging; 15-18 April
2004; Arlington, VA, USA; 2004. 2: pp. 1155-1158

[5] Situ N, Yuan X, Chen J, Zouridakis G. Malignant melanoma detection by


Bag-of-features classification. In: 30th Annual International Conference of the
IEEE Engineering in Medicine and Biology Society; 20-25 August 2008;
Vancouver, Canada; 2008. pp. 3110-3113
[6] Aswin RB, Abdul Jaleel J, Salim S. Implementation of ANN classifier using
MATLAB for skin cancer detection. In: International Conference on Mobility in
Computing-ICMiC13; 17-18 December 2013; India; 2013. pp. 87-94

[7] Cheerla N, Frazier D. Automatic melanoma detection using multi-stage neural


networks. International Journal of Innovative Research in Science, Engineering
and Technology. 2014;3(2):9164-9183

[8] L.G.Kabari,F.S.Bakpo ”Diagnosing Skin Diseases Using an Artificial Neural


Network ”,13 March 2009 in IEEE

[9] Tanzila Saba & Muhammad Attique Khan & Amjad Rehman & Souad Larabi
Marie-Sainte Region Extraction and Classification of Skin Cancer: A
Heterogeneous framework of Deep CNN Features Fusion and Reduction, ​Journal
of Medical Systems, September 2019

[10] Nikita Raut, Aayush Shah, Shail Vira, Harmit Sampat A Study on Different
Techniques for Skin Cancer Detection, September 2018. International Research
Journal of Engineering and Technology.

[11] Arneesh Aima, Akhilesh Kumar Sharma Predictive approach for Melanoma
Skin Cancer Detection using CNN, International Conference on Sustainable
Computing in Science, Technology & Management (SUSCOM-2019).

You might also like